Gesture recognition with Kinect
- Industry
- Human-computer interaction
- Type of work
- Sequence machine learning
- Period
- 2012
Architecture
Goal
Recognize hand gestures from the skeleton Kinect provides. Built for the User Experience course at UAI.
What I did
- Captured the hand trajectory from the Kinect skeleton.
- Quantized each position with K-means to turn movement into a symbol sequence.
- Trained one hidden Markov model per gesture and classified by maximum likelihood.